How hot is too hot?

Q

QuesoGrande

I have an Intel NUC based machine with 4 CPUs. It seems to be running hotter than I would like, but I have been away from matters like this for a while. Therefore I seek current thinking on the subject.

I have loaded the coretemp kernel module and the hwstat package, and all went well. When I run hwstat it is showing the CPUs to be running about 50.0 to 54.0 degrees C. Personally, I consider that a bit too hot. But what is the current day thinking about CPU temperature these days. Is that really too hot, or can it rise yet a bit before causing trouble? How about before shortening CPU life?

I have looked at forum posts and done a number of web searches, but cannot seem to find a consensus about the matter. Most references seem to indicate 40's C as fairly normal. But what of 50's C?

Advice appreciated!!!

QG
 
The T-Junction of the Intel CPUs in your NUC is likely somewhere around 100degC; your temps of 54degC are well below this and suggest the CPUs are nowhere near their operating maximum.
When they reach their T-Junction, they will automatically reduce their frequency to stay below this limit. They will not self-combust or destroy themselves through heat; they have very different upper temperature limits to human beings, for example.
 
QuesoGrande said:
I have looked at forum posts and done a number of web searches, but cannot seem to find a consensus about the matter. Most references seem to indicate 40's C as fairly normal. But what of 50's C?
Click to expand...
I'd monitor CPU frequency at high temps. With my i5-8400H, it keeps turbo 4.1GHz at average temps/load, but it drops to 3GHz slowly as temps rise. My laptop's cooling seems to keep it around 3GHz @ 90C.

I think 90C is a safe general limit for Intel, and I recall 50-60C for AMD Phenom II stuff, but ideally check the specific CPU specs.
 
  • Like
Reactions: bjs
QuesoGrande said:
How about before shortening CPU life?
Click to expand...
Besides not using it everything shortens its lifespan. Nothing lasts forever. Also semiconductors wear-out. But in all non cases of a warranty claim of normal usage within specs the to be expected lifespan is way more than sufficient.
In the attic I have an AMD Phenom II X4 980 running for almost 18 years with its included box cooler; others here may say: "Ludicrous! That's nothing! My...." - so, enough.

Question is, which temperature leads to intolerable accelerated shortening of life span.
Answer: In theory it's somewhere above ~90°C. In practice it doesn't really matter, as long as a sufficient cooling system is used. That's given in any case when using at least the recommended, or the included cooling device; mostly just the bare minimum, but fully sufficient, as long as you don't do any stunts like overclocking. While most self-assembled machines have notedly more cooling reserves, primarily for getting a quieter machine, but after all it's more cooling power.
And, of course, the cooler is properly installed. Most people I saw use way too much thermal paste. Even pros on YT videos seem to misbelief:"The more, the better" - wrong! Even the best thermal pastes ("liquid metal") do have a thermal conductance magnitudes lower as the direct contact bare metal to bare metal. The thermal paste is meant to close unavoidable air gaps, but the cooler must not be separated from the CPU's surface swimming on a film of it.
On laptops however the cooling capabilities of some (cheap) models may a bit just under what you'd need for a many weeks lasting continous constant 100% full power operation. But that's what they are not to be meant for, are they?
However, serious damages may not happen as long as you stay within specs, since every today's CPU has an internal thermal emergency safety shut-down. Worst thing could happen - e.g. when the bare minimum cooler has collected too much dust over many years, and does not deliver sufficient cooling power at maximum continous load anymore - is your machine suddenly blacks out. Then you may be concerned about not saved work, or your filesystem's integrity, or other things the interrupt may cause, but not damages to the CPU.

So, bottom line:
Anything below ~90°C is nothing to be concerned of, especially not 55°C.
Anything up to 70...80°C is just normal usage temperature at medium load, if not using a total overkill cooling system, but a fully sufficient one.

But if you were just begging for compliments how cool your machine was, at light loads mine's at 27°C 😜
 
I thank all of you who replied. It basically confirms what I thought, but I still have an old-fashioned feeling about this. It seems that present day cpus will tolerate nearly 90 - 100 deg C without much degradation. Therefore my concern at 50-some C is not especially cause for alarm. Amazing what technology does while I am looking the other way. All that having been said, I agree with bgavin that generally heat is the enemy of electronics. I would expand to all things electrical as well. Each has its ranges of acceptability. Just because of that I am pondering a couple of schemes for adding a fan to the mix. But it seems less urgent than it was before.

And as a side note to Maturin, while I wasn't looking for bragging rights, it seems that your range of temperature and mine are at least a little similar. My TZONE1 runs at about 25 to 28 C depending on loading at the time. Gives me a good feeling to someone else in the same range!! I think you gave good advice about the ranges to be and not be concerned about.

I think I consider this closed.

Thanks to all!!!

QG
 
I backed up a bit simply to convince you there's nothing you need to be concerned about, cause it's not.
And it seems it worked. :cool:

"I agree with bgavin that generally heat is the enemy of electronics. I would expand to all things electrical as well."
Heat is the enemy of everything. An electric conductor is not directly destroyed by too high voltage, or current, but because of the heat the current causes, which can be caused by too much voltage. Also bearings, gears, engines are suffering when not properly cooled. (Most) Proteins being destroyed above ~50°C. Extremely high temperatures even destroy atoms. 🥸 {weisenheimer off}
 
Maturin said:
..."I agree with bgavin that generally heat is the enemy of electronics. I would expand to all things electrical as well."
Heat is the enemy of everything. An electric conductor is not directly destroyed by too high voltage, or current, but because of the heat the current causes, which can be caused by too much voltage. Also bearings, gears, engines are suffering when not properly cooled.
Click to expand...

Even the back of fans might benefits from a fan :p

I've had two TP-Link AX1500 routers out-the-box drop wifi with high-load, until I took the top cover off and put a fan on it. I have a crate full of PC fans and some 5-12V barrel adapters lying around, so anything that looks like it'd benefit from a fan can get one :p
 
When I decommission hardware I always dismount fans and coolers, and keep them.
The coolers are only useful for tinkering in the lab, only.
But it's always good to have some fans handy.

To become a bit weisenheimer on that topic again:
The fans don't actually cool, they accelerate the distribution of the heat to the surrounding fluid (air), which distribute the heat even more, and tranport it away to where it doesn't hurt anymore, for the cooler can be made (a lot) smaller.
But as long as you cannot place your machine somewhere soundproof the trick ain't to realize enough cooling power. To realize as silent cooling as possible is - or even better trying to prevent cooling by fans at all.
Don't underestimate the effect on your sanity if you're subjected for years to the continuous noise of fans every day for many hours.🧐

Dusty fan blades are the main source of noise.
Tip: Lacquer the fan blades with non-stick coating makes them get dusty a bit slower.
Tip: Adjust your BIOS settings (if possible) to the fans run at low RPM when the CPU is on light loads only.
 
You must log in or register to reply here.
Back
Top